Research map / Point clouds / Shape and surface completion
HSurf-Net: Normal Estimation for 3D Point Clouds by Learning Hyper Surfaces
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use hyper surface fitting instead of 1-order jet fitting (ablation) | HSurf-Net · PCPNet average | angular RMSE: 12.12 -> 10.11 | Helped |
| use hyper surface fitting instead of 2-order jet fitting (ablation) | HSurf-Net · PCPNet average | angular RMSE: 12.75 -> 10.11 | Helped |
| use residual blocks instead of concatenation in fitting module (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.30 -> 10.11 | Helped |
| use Space Transformation instead of PointNet-based aggregation (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.61 -> 10.11 | Helped |
| add multi-scale Global Shift Layer vs fixed scale (ablation) | HSurf-Net · PCPNet average | angular RMSE: 11.17 -> 10.11 | Helped |
| use learned relative position encoding instead of traditional encoding (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.29 -> 10.11 | Helped |
| add relative position encoding vs raw coordinates (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.42 -> 10.11 | Helped |
| add point weights in output module (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.71 -> 10.11 | Helped |
| add neighborhood consistency loss L_con (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.11 -> 10.29 | No clear effect |
| replace sin loss with MSE loss (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.11 -> 15.45 | Hurt |
| change patch size from 700 to 600 or 800 (ablation) | HSurf-Net · PCPNet average | angular RMSE: 10.11 vs 10.16 (600), 10.15 (800) | No clear effect |
Closest papers by meaning
- SHS-Net: Learning Signed Hyper Surfaces for Oriented Normal Estimation of Point Clouds (CVPR 2023 · similarity 0.93)
- An Explicit Frame Construction for Normalizing 3D Point Clouds (ICML 2024 · similarity 0.88)
- A Lightweight UDF Learning Framework for 3D Reconstruction Based on Local Shape Functions (CVPR 2025 · similarity 0.87)
- LiSu: A Dataset and Method for LiDAR Surface Normal Estimation (CVPR 2025 · similarity 0.86)
- CMG-Net: Robust Normal Estimation for Point Clouds via Chamfer Normal Distance and Multi-Scale Geometry (AAAI 2024 · similarity 0.86)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.